Verification Standards

Goal: This site publishes only conclusions that have passed verification. Conclusions that do not pass are published honestly as negative results: "what I tested that did not work" is just as important as "what works."

What it does: Distinguish whether a signal is a real edge or a statistical illusion. Most "effective signals" in the market die in a few classic traps: backtesting only on stocks that survived, accidentally using future information, getting inflated by a few stocks that exploded, or forgetting to subtract trading costs. This site's verification system exists to block those traps.

Requirements -- Before publication, any quantitative conclusion on this site must satisfy:

• Tested on a complete historical sample that includes delisted stocks
• Signal calculations use only information available at the time
• Evaluated using net returns after trading costs, and compared against a "buy anything" baseline
• Statistical significance is appropriately corrected
• Major conclusions receive independent review

Conclusions that do not meet the requirements will not appear here; if a published conclusion is overturned, it will be publicly corrected. The internal design of the verification system is not publicly disclosed.